Fix dag run type enum query for mysqldb driver

[houqp]

By default sqlalchemy pass query params as is to db dialect drivers for
query execution. This causes inconsistent behavior of query param
evaluation between different db drivers. For example, MySQLdb will
convert DagRunType.SCHEDULED to string 'DagRunType.SCHEDULED'
instead of string 'scheduled'.

To keep the behavior consistent across DB dialects, this patch
introduces a new enum aware sqlalchemy type based of
sqlalchemy.types.String to do the query param conversion within
sqlalchemy's ORM layer before passing the value down to individual DB
driver.

see #11621 for relevant
discussions.

[ashb]

See comments

[ashb]

Why didn't mysql unit tests fail before?

[github-actions]

The Workflow run is cancelling this PR. It has some failed jobs matching ^Pylint$,^Static checks,^Build docs$,^Spell check docs$,^Backport packages$,^Provider packages,^Checks: Helm tests$,^Test OpenAPI*.

[houqp]

@ashb because when running the tests, run_type are being written as string 'DagRunType.SCHEDULED' into the db, so when it's queried using the same value, tests passed. We ran into this issue during upgrade because all of our old db rows have the column populated as string 'scheduled', which lead to crashes due to queries returning empty result.

All existing 1.x mysql users using mysqlclient driver will run into this issue after performing the 2.0 upgrade. New 2.0 mysql users might already have db populated with 'DagRunType.SCHEDULED', which needs manual fix up.

I will prepare a test case to guard against this regression going forward.

[ashb]

@ashb because when running the tests, run_type are being written as string 'DagRunType.SCHEDULED' into the db, so when it's queried using the same value, tests passed. We ran into this issue during upgrade because all of our old db rows have the column populated as string 'scheduled', which lead to crashes due to queries returning empty result.

All existing 1.x mysql users using mysqlclient driver will run into this issue after performing the 2.0 upgrade. New 2.0 mysql users might already have db populated with 'DagRunType.SCHEDULED', which needs manual fix up.

I will prepare a test case to guard against this regression going forward.

Oooh. Fuuu.

[ashb]

(can't seem to comment without leaving a review anymore on Android app)

[potiuk]

All existing 1.x mysql users using mysqlclient driver will run into this issue after performing the 2.0 upgrade. New 2.0 mysql users might already have db populated with 'DagRunType.SCHEDULED', which needs manual fix up.

Not good :(

[kaxil]

I am surprised that this slipped in alpha/beta and rcs

[ashb]

?

[github-actions]

The Workflow run is cancelling this PR. It has some failed jobs matching ^Pylint$,^Static checks,^Build docs$,^Spell check docs$,^Backport packages$,^Provider packages,^Checks: Helm tests$,^Test OpenAPI*.

[github-actions]

The Workflow run is cancelling this PR. It has some failed jobs matching ^Pylint$,^Static checks,^Build docs$,^Spell check docs$,^Backport packages$,^Provider packages,^Checks: Helm tests$,^Test OpenAPI*.

[kaxil]

Just rebased, let's merge this one once it passes CI

Stale Review